Adjusted quantized tolerance for graph as it was too low (pytorch#12653)

### Summary
Qtol was previously too low for the graph and it would break the CI. The
error is expected with the quantized graphs.

### Test plan
Re-ran all tests with new qtol.

Author: leafs1

backends/xnnpack/test/ops/test_to_copy.py

@@ -110,4 +110,4 @@ def test_quantized_to_copy(self):
                 "executorch_exir_dialects_edge__ops_aten__to_copy_default",
                 "executorch_exir_dialects_edge__ops_quantized_decomposed_quantize_per_tensor_default",
             ]
-        ).to_executorch().serialize().run_method_and_compare_outputs(qtol=0.01)
+        ).to_executorch().serialize().run_method_and_compare_outputs(qtol=1)